Prediction And Analysis Of Milling Deformation For Thin-walled Workpiece Based On Optimized Neural Network

With the level of manufacturing improves continuously in our country, thin-walled workpiece is widely applied in the national equipment of aerospace, weapon, national defense and so on. The shape of thin-walled workpiece is more complex, and it required higher machining accuracy of the surface. In the milling process, the thin-walled workpiece is easy to deform due to the cutting force and cutting heat etc. This could have an impact on the machining accuracy, working performance of workpiece and even reduce the machining efficiency. Therefore, Studying machining deformation of thin-walled workpiece during machining and effectively controlling machining deformation are very meaningful to aerospace, weapon manufacturing etc.During machining thin-walled workpiece, the structure of tool has a great influence on cutting force and cutting heat. It could result in different machining deformation, even affect the machining efficiency. To research the influence of tool structure on machining deformation, this paper reasonably determines the geometric parameters of its structure for machining thin-walled workpiece so that machining deformation could be effectively controlled. The main works of this paper are listed as following:(1)Realize development of the parametric parts library of tool. Based on the secondary development of UG, the parametric design idea and database technology are used for the development of the parametric parts library of tool. A friendly user interface of the parametric parts library of tool is established. It can provide three-dimensional tool model for the milling simulation of thin-walled workpiece.(2) Study the key technologies of simulation. By using the parametric parts library of tool, the three-dimensional geometry model of tool can be created, and then imported to ABAQUS finite elements software. Based on the coupled thermo-mechanical action, the milling process of aluminum alloy 7050- T7451 thin-walled workpiece is simulated. Compared with experimental data, the simulation results demonstrate that the simulation finite element model for the milling process of thin-walled workpiece is reasonable and valid.(3) On the basis of the simulation finite element model for the milling process of thin-walled workpiece, the nonlinear mapping relationship between geometric parameters of tool and machining deformation is obtained by establishing the model of predicting machining deformation based on genetic evolution neural network. BP network’s initial weights and threshold are chosen as the design variables, and then establish the optimization model of initial weights and threshold. The Genetic Algorithm which has a strong ability of global optimization is utilized to solve the optimal initial weights and threshold. Finally, the geometry parameters(rake angle, relief angle, and helix angle) of tool are taken as the design variables, and an optimal model with the objective of minimizing the maximum deformation is developed. The Genetic Algorithm is used to solve the optimal problem, and a set of optimized rake angle, relief angle and helix angle are consequently obtained. This paper can provide a basic theory of design and selection of milling tool for the thin-walled workpiece so that the machining deformation of thin-walled workpiece in milling process can be effectively controlled.